How long does it take to learn a language?

One institution has been measuring this for decades and publishes the answer: the US State Department, which has to budget the time before it can post a diplomat. Its table is where almost every number you have read comes from. It is also routinely quoted in a way that makes it mean something it does not, because the headline figure counts only classroom hours and the same page says a typical training week carries seventeen more hours on top.

What is the short answer?

Between roughly 550 and 2,200 classroom hours, depending on how far the language sits from English, and that is before self-study. The US Department of State's Foreign Service Institute sorts languages into four categories and publishes a week count and an hour count for each: Category I at 24 to 30 weeks and 552 to 690 class hours, Category II at about 36 weeks and 828 hours, Category III at about 44 weeks and 1,012 hours, and Category IV at 88 weeks and 2,200 hours. Those are the only official training budgets published by anyone, which is why every article on this subject traces back to them.

The part that usually gets dropped is on the same page, one sentence above the table. “A typical week is 23 hours per week in class and 17 hours of self-study.” Multiply the weeks by 40 rather than 23 and the real commitment appears: 1,760 hours for a Category III language, not 1,012, and 3,520 hours for Japanese or Arabic, not 2,200. The published figure is not the time the language takes. It is the time the classroom takes.

The Foreign Service Institute budgets 1,012 class hours over about 44 weeks for a Category III language such as Hebrew, Russian or Turkish, and states that a typical week is 23 hours in class plus 17 hours of self-study, which puts the full commitment at roughly 1,760 hours (US Department of State, read 11 September 2026).

What are the four FSI language categories, and what does each cost?

The categories are a measure of distance from English, not of a language's complexity in the abstract. The page names 59 languages across the four groups and says in the same breath that it trains 60, and it labels the Category III list explicitly as “not exhaustive”. The distribution is lopsided: nine languages are close to English, five sit in the second tier, forty are classed as hard, and exactly five are called super-hard.

One arithmetic detail is worth knowing before you quote the table. For Categories I, II and III the hour figures are exactly the week counts multiplied by the 23 class hours a week the page states: 23 times 24 is 552, 23 times 30 is 690, 23 times 36 is 828, and 23 times 44 is 1,012. Category IV does not follow that rule. Eighty-eight weeks at 23 hours would be 2,024, while the page prints 2,200, which implies 25 hours a week. Whichever figure you use, check which one your source used.

Why do some sources say 1,100 hours and others say 1,012?

Because two editions of the same table are in circulation and they disagree. The current State Department page gives Category I as 552 to 690 class hours, Category II as 828 and Category III as 1,012, and targets “an integrated score of 3 (Speaking + Listening)” on the ILR scale. An earlier edition gave the same week counts converted at 25 hours a week instead of 23, producing 600 to 750, 900 and 1,100, and targeted Speaking-3 and Reading-3 separately. The week counts, which are the real finding, never changed.

This matters more than a rounding argument, because the older numbers are the ones that propagated. If an article gives you 1,100 hours for Russian or 600 hours for Spanish without naming a retrieval date, it is quoting the superseded edition. The safest way to cite this table is to give the week count, which both editions agree on, and to say which hour conversion you used.

What proficiency do those hours actually buy?

Not what most people mean by fluent, and this is the single largest source of confusion about the FSI figures. The target is level 3 on the Interagency Language Roundtable scale, which runs from 0 to 5. The ILR defines Speaking 3, General Professional Proficiency, as being able to “participate effectively in most formal and informal conversations in practical, social and professional topics”, with language used acceptably but with noticeable imperfections, where “errors virtually never interfere with understanding and rarely disturb the native speaker”. That is the standard for conducting a meeting or defending a policy position in the language.

The level below it is closer to what a person usually has in mind. ILR Speaking 2, Limited Working Proficiency, is defined as being “able to satisfy routine social demands and limited work requirements”, handling high-frequency conversation with confidence but not facility, and the scale says plainly that at this level “errors are frequent”. If your goal is to hold a real conversation, make friends and get through a working day, you are aiming at a band the FSI budget overshoots. Nobody publishes an hour figure for ILR 2, but it is unambiguously fewer hours than the number you have been quoted.

The FSI hour budgets target ILR level 3 of 5, General Professional Proficiency, which the Interagency Language Roundtable defines as speaking where “errors virtually never interfere with understanding”. The level most people mean by conversational is ILR 2, where the same scale says “errors are frequent”.

How many hours to reach B1, B2 or C1 in English?

There is a second official hour table, built on a different scale and a different population, and it is the more useful one if you are learning English rather than being posted abroad by a government. Cambridge University Press and Assessment states that “it takes approximately 200 guided learning hours for a language learner to progress from one level of the Common European Framework of Reference (CEFR) to the next”, and publishes cumulative totals from complete beginner for each level.

Read the two tables together and they are not in conflict, because they finish in different places. Cambridge puts C1, the level of its C1 Advanced exam, at 700 to 800 guided hours from complete beginner. The FSI budgets 1,012 class hours to reach ILR 3 in a Category III language, and 552 to 690 for a language close to English. Those ranges overlap. The FSI number looks larger mainly because its languages are further from the learner's own and because it counts the classroom hours of a full-time programme.

How long will it take me, on my actual schedule?

This is the question behind the question, and it is the one the published tables never answer, because the FSI does not have the problem you have. Its students study all day. You have an evening. The hours do not change; the calendar does, and the calendar is the entire story. Below is the same published class-hour budget divided by four ordinary schedules, with no allowance for missed days.

The same 1,012 class hours the State Department budgets for Hebrew or Russian is about ten months at the FSI's full-time pace, about two years and nine months at an hour a day, and about five and a half years at half an hour a day.

The gap between the first row and the last is the finding. It is the same language and the same hour budget; only the density of the schedule changed. That is also why the honest answer to “how long does it take” is a counter-question: how many hours a week, and towards which level. Anybody who answers in months without asking either is quoting a number they have not checked.

Does the type of hour matter, or only the count?

The published tables count hours, not kinds of hour, and that is their real limitation. The FSI describes a week of 23 hours in class plus 17 of self-study, which means that in the programme these figures come from, well over half the timetable is classroom time rather than private study, which is time spent producing the language out loud and being corrected on the spot. A self-study hour spent reading, watching or tapping through an app is counted by nobody as equivalent, and the tables never claim it is.

This is where most adult learners lose years without noticing. Input accumulates easily, because it is comfortable and available. Output does not, because it requires a listener and a willingness to be wrong in front of one. A learner who has watched hundreds of hours of English television and never held a conversation has a large hour count and a small speaking level, and no table will explain the discrepancy, because the tables assume the FSI's mix of classroom hours rather than a person's actual one.

Frequently asked questions

Related research

The same FSI table read from the other direction, for one language and one country, is what learning Hebrew costs an American, which pairs the hour budget with the Modern Language Association enrollment census. For the gap between the hours a country puts in and the speaking it gets out, every country reads English better than it speaks it puts seven countries side by side, and how many people speak English audits what each official source actually counts. A Japanese edition of the hours argument, built on this table and the standard lesson counts in Japanese school regulations, is at 英語習得にかかる時間. The same table read as a ranking, all 59 languages by category, is easiest languages to learn for English speakers, and the question turned around, how hard English is for speakers of each of those languages, is is English hard to learn.